Sei sulla pagina 1di 31
To be presented at the 3i7tk meeting of the American Institute of Electrical Engineers, New
To
be presented at the 3i7tk meeting of the
American Institute of Electrical Engineers,
New
York, January 14, 1016.
Copyright 1916. By A. I. E. E.
(Subject to final revision for the Transactions.)
OUTLINE OF THEORY OF IMPULSE CURRENTS
BY
CHARLES P. STEINMETZ
ABSTRACT OF PAPER
In
Part I it is shown how, from the integral of the general
differential equation of the electric circuit, which has been dis-
cussed in a previous paper, all the types of electric currents are
derived as special cases, corresponding to particular values of the
integration constants.
The
equations of the circuits with massed constants, th»"t
is,
the usual electric circuits, are derived by substituting zero
for
the (electrical) length of the circuit.
Besides the general transients, discussed in a previous
paper, three main classes of currents are shown to exist,
corresponding to different values of the time exponent b:
The
alternating currents, corresponding to b = imaginary,
whioh are the useful currents of our electric circuits.
The
impulse currents, corresponding to real values of ft,
which may be called harmful currents of our electric circuits.
And,
as limit case, for b - 0, the continuous-current circuit
with
distributed constants.
The
last case, a continuous current in a circuit with dis-
tributed resistance and leakage, is discussed, and it is shown
that
such continuous-current circuit has many features which are
usually considered as typical of alternating-current wave trans-
mission. It consists of a main current and a return current;
complete reflection occurs at the end of the circuit; partial
reflection at a transition point; a surge resistance exists, which,
connected to the circuit, passes the current without reflection.
In
Part II an outline of the theory of impulse currents is
given. They comprise two classes, the non-periodic and the
periodic impulse currents. The equations of both are given
in
different form, by exponential and by hyperbolic or trigono-
metric functions. Their characteristics are:
Impulse current and voltage may be resolved into a main
wave
and a return wave. The return wave is displaced from the
main
wave in time and in position. A time displacement exists
between the two current waves and their corresponding voltage
waves. This time displacement may be a lag of the current be-
hind
the voltage, or a lead, depending on the circuit constants.
In
the periodic impulse currents, the displacement between main
wave
and return wave is represented by a position angle, and
the
two current waves are in quadrature in position, to their
respective voltage waves.
A
few special cases are discussed.
Manuscript of this paper was received December 10, 1915.
1
Generated on 2012-03-09 21:16 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
2 STEINMETZ: IMPULSE CURRENTS [Jan. 14 I. TYPES OF CURRENT A. GENERAL IF r =
2 STEINMETZ: IMPULSE CURRENTS [Jan. 14
I.
TYPES OF CURRENT
A.
GENERAL
IF
r = resistance
g
= shunted conductance
L
= inductance
C
= capacity
per
unit length of any circuit or section of a circuit, then in any
line
element dl of the circuit, the voltage consumed is
d
e . T di ,.,.
L
-T- (1)
dl
dt
and
the current consumed is
—
= e + C — (2)
Differentiating (1) over dt, and (2) over dl, and combining
gives
Integrating,
i
= A 6-" e-6' (4)
e = -— A e-" r-4' (5)
= \/bL- r Af-«e-t,
= z i
where
.
=
V*4^ (6)
Generated on 2012-03-09 21:16 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] STEINMETZ: IMPULSE CURRENTS 3 is the surge impedance or natural impedance of the circuit,
1916] STEINMETZ: IMPULSE CURRENTS 3
is
the surge impedance or natural impedance of the circuit, and
a
and b are related by the equation
a2 = PLC - b (rC + gL) + rg (7)
= (bL - r) (bC - R)
The
general solution then is a sum of such terms (4) and (6).
These equations must represent every existing electric cir-
cuit,
and every circuit which can be imagined, from the lightning
discharge to the house bell, and from the alternating-current
transmission line to the telephone circuit, with the only limita-
tion,
that r, g, L, C are constant within the range of the currents
and
voltages considered.
The
difference between all electric circuits thus merely consists
in
the difference of the constants A, a and b, and the difference
in
the length / of the circuit.
If
I = 0, that is, the length of the circuit is negligible compared
with
the rate of change of i or e, (4) and (6) give the equations
of
all the circuits with massed constants, otherwise we get the
equations of the circuits with distributed constants.
In
general, a and b are general numbers, related to each other
as
in (7). Choosing b as the independent constant,
b
= 0 gives the continuous currents,
b
= real gives the impulse currents.
b
= imaginary gives the alternating currents.
b
= general or complex imaginary gives the general transients.
b
= 0 or imaginary thus gives the permanents, continuous
current and alternating current.
b
= real or general gives the transients, impulse currents and
general transients.
Thus, while the continuous currents represent a limiting case,
the
alternating currents and the impulse currents are two co-
ordinate classes of currents. While the alternating currents are
the
useful currents of our electric systems, the impulse currents
may
be said to be the harmful currents in our systems, as many
of
the disturbances and troubles in electric systems are caused
by
them. i
The
theory of alternating currents is discussed in numerous
publications, but little systematic study has been made of the
impulse currents, and their general theory thus will be given in
the
following, and also that of the limiting case of the continuous
Generated on 2012-03-09 21:16 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
4 STEINMETZ: IMPULSE CURRENTS [Jan. 14 currents in a general circuit of distributed constants. The
4 STEINMETZ: IMPULSE CURRENTS [Jan. 14
currents in a general circuit of distributed constants. The theory
of
the general transients has been outlined in a previous paper.*
B.
CIRCUITS WITH MASSED CONSTANTS
Substituting in (4) and (5):
e
= 0 when / = 0
and
considering that, by equation (7), a can have two values, for
every value of b, + a and — a, we have,
i
= A e-6' {€-"' + e+^1
(8)
e
= bL~ T A «-* {e-"- e+"}
a
<
Assuming now / as infinitely small, and substituting
e
-"' = 1 ± a / (9)
in
(8), gives, as the general equation of the circuit with massed
constants,
i
= B e-6'
(10)
e
= (r» - 6L0) B 6~>"
where
B
= 2 A
TO
= / r = total resistance of circuit,
Lo
= IL = total inductance of circuit.
Substituting b = 0 in (10), gives
i
= B
(11)
e
= r0B
as
the equation of the continuous-current circuit.
For
b = real (10), the equations of the inductive discharges are
i
= B e-<"
e = (r0 - bL0) B «-"
= r0 i + L0 ~- (12)
*A.
I. E. E. TRANSACTIONS 1908, page 1231, and more completely
in
Section IV of " Transient Phenomena."
Generated on 2012-03-09 21:17 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] STEINMETZ: IMPULSE CURRENTS 5 Substituting b = ±.jc gives as the equation of the
1916] STEINMETZ: IMPULSE CURRENTS 5
Substituting b = ±.jc gives as the equation of the alternating-
current circuit,
i
= Ci cos ct + Ct sin ct
(13)
e
= (TO Ci — c Lo Ct) cos ct + (ro Ct + clo Ci) sin ct
where
(,\ = X>1 ~T~ .D?
(14)
C2
= j (B, - Bt)
C. CONTINUOUS-CURRENT CIRCUIT WITH DISTRIBUTED
CONSTANTS
b
= 0
From (7) we have
a
= ± Vrg (15)
and
from (6),
z
= ± V- (16)
substituting (16) and (16) in (4) and (6) gives
(17)
*
/7
e
= V - \Ai «-' ^ - At e+lvS (18)
S
These equations do not contain L and C, that is, inductance
and
capacity have no effect on the permanent continuous-current
circuit with distributed constants, but only resistance and con-
ductance, that is, leakage.
Equations (17) and (18) are the equations of a direct-current
circuit having distributed leakage, such as a metallic conductor
submerged in water, or the current flow in the armor of a cable
laid
in the ground, or the current flow in the rail return of the
direct-current railway, etc.
r
is the series resistance per unit length, g the shunted or leak-
age
conductance per unit length of circuit.
Generated on 2012-03-09 21:17 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
6 STEINMETZ: IMPULSE CURRENTS [Jan. 14 Where the leakage conductance is not uniformly distributed, but
6 STEINMETZ: IMPULSE CURRENTS [Jan. 14
Where the leakage conductance is not uniformly distributed,
but
varies, the numerical values in (18) change wherever the
circuit constants change, just as would be the case if the resist-
ance
r of the conductor changed. If the leakage conductance
g
is not uniformly distributed, but localized periodically in space
—as at the ties of the railroad track, — when dealing with a suffi-
cient
circuit length, the assumption of uniformity would be
justified as an approximation.
(a)
If the conductor is of infinite length — that is, of such great
length, that the current which reaches the end of the conductor,
is
negligible compared with the current entering the conductor,
we
have
A-,
= 0
This
gives
4/7
(19)
e
= A V — e-'V'i
or
e=\/^i
(20)
that
is, a conductor of infinite length (or very great length) of
series resistance r and shunted conductance g, has the effective
resistance r' = 'V — .
g
It
is interesting to note, that at a change of r or of g the effect-
ive
resistance r', and thus the current flowing into the conductor
at
constant impressed voltage, or the voltage consumed at con-
stant
current, changes much less than r or g.
(b)
If the circuit is open at / = lo, we have
i
= Al e-'« + A,. e+i'~* = Q
Hence, if
A
= Ai e~l'v^ = - A* f+''vn
we
have
e
= A V --
Generated on 2012-03-09 21:17 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] STEINMETZ: IMPULSE CURRENTS 7 (c) If the circuit is closed upon itself at I
1916] STEINMETZ: IMPULSE CURRENTS 7
(c) If the circuit is closed upon itself at I = /0> we have
e = V— {A, e-fc^I- A*
= 0
Hence, if
A
= A i
we
have
=
A
=
A V — {e+(fa-o ^ - e- ('«-') ^
(22)
If,
in (22), /0 = 0, that is, the circuit is closed at the starting
point, we have
t
= A { e-/v^ + e+'^j
\fr
_i ^~ +lvr,
'
g
or,
counting the distance in opposite direction, that is, changing
the
sign of /
«
= A (e+IX/v« + e-'v/" }
e
= X V- |e+'v''» - e-'^!
(23)
If
the circuit, at / = /0, is closed by a resistance r0, we have
=
''o
hence.
Generated on 2012-03-09 21:17 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
STEINMETZ: IMPULSE CURRENTS [Jan. 14 A! €-* or, A/I- r» A, = .4,6-2 (24) —
STEINMETZ: IMPULSE CURRENTS
[Jan. 14
A!
€-*
or,
A/I-
r»
A,
= .4,6-2
(24)
—
+
g
t
= A
r.-V-
r»
JL
,-(2/0-0
(25)
+
:
- (2 I, -
These equations (23) and (26) can be written in various dif-
ferent forms. They are interesting in showing in a direct-current
circuit, features which usually are considered as characteristic
of
alternating currents, that is, of wave motion.
The
first term of (23) or (26) is the out-flowing or main current
or
voltage, respectively; the second term is the reflected one.
At
the end of the circuit with distributed constants, reflection
occurs at the resistance r0.
VI,
If
r0 > V —, the coefficient of the second term is positive, and
o
partial reflection of current occurs, while the return voltage adds
itself
to the incoming voltage.
If
r0 < v —, reflection of voltage
occurs, while the return
current adds itself to the incoming current.
If
TO = v —, the second term vanishes, and the equations
(26) become those of (19), of an infinitely long conductor. That
Generated on 2012-03-09 21:18 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] STEINMETZ: IMPULSE CURRENTS 9 A resistance r0, equal to the effective resistance (surge imped-
1916] STEINMETZ: IMPULSE CURRENTS 9
A
resistance r0, equal to the effective resistance (surge imped-
ance) v — of a direct-current circuit of distributed constants,
g
passes current and voltage without reflection. A higher resist-
ance
ro partially reflects the voltage—completely so for ro = °°,
or
open circuit. A lower resistance ro partially reflects the cur-
rent—completely so for ro = 0, or short circuit.
—
thus takes in direct-current circuits the same position
g
as
the surge impedance in alternating-current or transient circuits.
D. ALTERNATING-CURRENT CIRCUITS WITH DISTRIBUTED
CONSTANTS
b
= ±jq
by
equation (7), a then becomes a general number: ± (h ± j k),
and
the corresponding values of b and a exist:
b = + j q, a = + h — j k
- jq + h +jk
- jq - h - j k
+
jq - h + jk
Substituting these in equation (4), and substituting for the
exponentials with imaginary exponent the trigonometric expres-
sion,
gives
i
= t~u \Bi cos (kl - qt) + Bt sin (k I- qt)\
+
e+u \B3 cos (kl + qt) + B< sin (kl + qt)\ (26)
where
BI
= A\ + A« B3 = A3 + A\
Resolving now the trigonometric expression of (26) into expres-
sions
of single angles, and eliminating the function of time by the
introduction of the vector notation,
Bt
cos qt — BI sin qt = B\ — j B* = A\.
Bt
cos qt + 5i sin qt = 52 + j BI = j A\.
B3
cos qt + B^ sin qt = B3 + j B^ — —At.
B^
cos qt — B3 sin qt = B^ — j B3 = j A^.
Generated on 2012-03-09 21:18 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
10 STEINMETZ: IMPULSE CURRENTS [Jan. 14 gives as the expression of the current, 7 =
10
STEINMETZ: IMPULSE CURRENTS [Jan. 14
gives as the expression of the current,
7
=
Ai €-*<(cos kl + jsin kl) - At e+w (cos kl - jsinkl) (27)
and
in the same manner, starting with equation (6),
£A
/ f • LI f i J i • • i J\ , J_t/ / i
—-
\f ( A . f ~ «* ( pn^ JP/ -J— 1 *i1 T1 *»/ I -i- -\ ,, f T ** (POQ JPI
v — _ i ^i i c ^v^Uo n* ~ / am KV) \^ <i J c ^\-uo Atr
- j sin */)) (28)
where
Z
= r + jqL
Y
= g+jqC (29)
q
= 27T/
These are the usual and well known equations of the alter-
nating-current transmission line in symbolic expressions.
II.
IMPULSE CURRENTS
A.
GENERAL
Impulse currents are characterized by the condition, that
the
time exponent b in equations (4) and (6) is real.
By
equation (7), to every value of b correspond two values
of
a, equal but of opposite sign:
±
a
By
the same equation, to every value of a correspond two
values of b:
b
= M± s (30)
where 5 = V m2 + -^ (31)
is
the energy transfer constant,
„
\
(32)
T-
\
is
the energy dissipation constant, and
'
* = H -ir ~ -r I
is
the distortion constant of the circuit.
Generated on 2012-03-09 21:18 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] STEIN METZ: IMPULSE CURRENTS 11 As b must be positive, it must be, by
1916] STEIN METZ: IMPULSE CURRENTS 11
As
b must be positive, it must be, by (30)
S2
< M2 (34)
Since u and b are real, by (30), 5 must be real, thus by (31),
a2
m2
+ j—^r must be real and positive.
JLrf
U
a2
As
fw2 is real, j—^r must be real, and must either be positive,
i
* (
a2
or,
if negative, — j— — must be less than m2.
/^
(
a
thus must be either real, or imaginary, but it can not be
complex imaginary.
This
gives two classes of impulse currents:
Non-periodic impulse currents: a real, a2 positive.
Periodic impulse currents: a imaginary, a2 negative, and
-
a2 < L C m2
The
terms " periodic " and " non-periodic " here refer to the
distribution in space, but not in time, since as function of time
the
impulse currents are always non-periodic.
The
relations between the constants thus are:
Non-periodic impulse currents:
a2
= positive
a
= ± h
=
V
+
r (36)
h
= VL C (52 - m2)
M2
> S2 > W2
and
corresponding values of a and b are :
a:
b:
+
h u- s
—
h u — s
- h u +s (36)
+ h u + s
Generated on 2012-03-09 21:19 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
12 STEINMETZ: IMPULSE CURRENTS [Jan. 14 Periodic impulse currents: a'2 = negative a = ±j
12
STEINMETZ: IMPULSE CURRENTS [Jan. 14
Periodic impulse currents:
a'2
= negative
a
= ±j k
k
= VLC (m*- **) (37)
jfe2
< ws L C
and
corresponding values of a and b are :
a:
b:
+
jk u- s
-
jk u- s (38)
+
jk u + s
-
jk u + s
B.
NON-PERIODIC IMPULSE CURRENTS
Substituting (35) and (36) in (4) and (6), and rearranging,
gives :
i
= «-"' 1/1, e-*'+" + At f+u+« + As e+w-« + At e-u-« }
-§-«- \cA> *-»+*-cAt
. J_ I +W ,_ J_ « rt I
, c ^3€ c ^4« ,
(39)
where
.
/ r _1_ «M
(40)
s
— m
These equations (39) can be simplified by shifting the zero
points of time and distance, by the substitution:
!
= Dt e-wi-"-
3 ± Di €-*i+"« (41)
=
4 ± Z?2 e+w' + s'-
=
c
= «+"• (42)
Generated on 2012-03-09 21:19 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] STEIN METZ: IMPULSE CURRENTS 13 Hence . *-!2F -•£•*-££ <»> and writing .'
1916] STEIN METZ: IMPULSE CURRENTS 13
Hence
. *-!2F -•£•*-££ <»>
and
writing .' for / — ti + ta
and
/ for / — l\
they
then assume the form:
-
D, [€+*+• <'-w± e-"-1 <'-">>]) (44)
«
= ^-V* {D, [«-*<+" ± « + «-«]
+
ZM«+W + S/ ± e-w-« ]}
or,
expressed in hyperbolic functions :
i
= e-1" \B! cosh [hi - st(t - ta)] - B-I cosh [hi + s(t -/o)]|
_
(45)
e
-V ~ «-"' {5, cosh [hi - st] + 52 cosh (hi + st\\
c/
or
the corresponding sinh functions, in case of the minus sign in
equation (44).
The
impulse thus is the combination of two single impulses of
the
form
e-»((e-W+J/ ± e+«-5l)
which move in opposite direction, the Z?i impulse towards rising
/:
—j— > 0, and the £>2 impulse towards decreasing I: — r— < 0.
ill
at
The
voltage impulse differs from the current impulse by the
factor v —p^- (the "surge impedance"), and by a time' dis-
\^
placement to. That is, in the general impulse, voltage e and the
current * are displaced in time.
t0
thus may be called the time displacement, or time lag of the
current impulse behind the voltage impulse.
Generated on 2012-03-09 21:19 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
14 STEIN METZ: IMPULSE CURRENTS [Jan. 14 ta is positive, that is, the current lags
14 STEIN METZ: IMPULSE CURRENTS [Jan. 14
ta
is positive, that is, the current lags behind the voltage im-
pulse, if in equation (43) the log is positive, that is, m is positive,
Y
!!
or:
—=— > -%-, that is, the resistance-inductance term pre-
ponderates.
Inversely, to is negative, and the current leads, or the voltage
impulse lags behind the current impulse, if m is negative, that is,
f
a
—j— < — %r- , or the capacity term preponderates.
If
g = 0, that is, no shunted conductances, the current impulse
always lags behind the voltage impulse.
If
m = 0, that is, — £— = -|r- , or -^— = —^- , to = 0, that
L
C g C
is,
the voltage impulse and the current impulse are in phase with
each
other, that is, there is no time displacement, and current
and
voltage impulses have at any time or at any space the same
shape; distortionless circuit, m therefore is called the distortion
constant of the circuit.
In
the individual impulse
f-»l(f- U + H ± f + U- jf) _ f-U e-(tt - J)l _j_ e+ Uf -(«+*)(
(46)
the
term e - «< is the attenuation of the impulse by the energy
dissipation in the circuit, that is, represents the rate at which the
impulse would die out by its energy dissipation.
The
first term: e~~("~s)/, dies out at a slower rate than given
by
the energy dissipation, that is, in this term, at any point /,
energy is supplied, is left behind by the passing impulse, and as the
result, this term decreases with increasing distance /, by the factor
€
• u ; inversely, the second term, e ' ("+j) ', dies out more rapidly
with
the time, than corresponds to the energy losses, that is, at any
point
/, this term abstracts energy and shifts it along the circuit,
and
thereby gives an increase of energy in the direction of pro-
pagation, by e+u. In other words, of the two terms of the im-
pulse, the one drops energy while moving along the line, and the
other
picks it up and carries it along.
The
terms e *'' thus represent the dropping and picking up of
energy with the time, the terms e*w the dropping and picking
up
of energy in space along the line. In distinction to u, which
Generated on 2012-03-09 21:19 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] STEINMETZ: IMPULSE CURRENTS 15 may be called the energy dissipation constant, s (and its
1916] STEINMETZ: IMPULSE CURRENTS 15
may
be called the energy dissipation constant, s (and its cor-
responding h) thus may be called the energy transfer constant of
the
impulse. The higher s is, the greater then is the rate of
energy transfer, that is, the steeper the wave front, and s thus
may
also be called the wave front constant of the impulse.
Substituting in equations (44), I — 0, gives the equation of
the
impulse in a circuit with massed constants:
i
= A e-* (e+J< db e--")
e
= B V ~- e-»< (e+'C-u ± «-'('-<•>)
where
A = DI — Z>2
B = D, + £>2.
Substituting in equation (39),
A,. = ± B e+w--*"+*
A, = ± B e-«,-«•-* (47)
A . — -i- R f 4*W, -4-s/i - x
^1
4
—• ^C jL/ C
c
= €+''• (42)
and
writing
,
r
,
,
i *
/
tor t—ti -\- /oH
s
I
tor I- /i
and
rearranging, gives
B\/X
e-«/ (e-«[e + S'd e-,. +* e+,(/-o (48)
C*
where
2
*
Generated on 2012-03-09 21:19 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
16 STEINMETZ: IMPULSE CURRENTS [Jan. 14 writing t lor t— h + t' I for
16
STEINMETZ: IMPULSE CURRENTS [Jan. 14
writing
t
lor t— h + t'
I
for l-h + —
n
the
substitution of (47) and (42), gives, after rearranging,
*
= B e-"' {e-'<»-« [t+M ± e-w] ± «+'«-*)
«
= sV-4- e—'le-1
±c
- » 1 I f +il
I
, - e J ^ c
(50)
where
0
-e
-
(51)
Equations (48) may be interpreted as showing two impulses,
the
main impulse, and the reflected impulse. The main im-
pulse, with f~u, decreases with increasing /, that is, progresses
towards rising /. The reflected impulse, with e+w, starts at
the
time /' after the start of the main impulse, and decreases
with
decreasing /, that is, progresses towards decreasing /.
The
two current impulses lag behind their voltage impulse
by
time /o-
Equation (50) shows the two component impulses, the first
one
dropping energy along its path, and thus decreasing with
the
time at a greater rate than corresponds to the energy dis-
sipation, and the second one, displaced in position from the
first
one by distance /', picking up the energy dropped by the
first
one.
The
current again lags behind the voltage by time /o.
The
distance displacement /' of the component impulse in
(50) is related to the time displacement t' of the two component
impulses in (48) t>y (49) and (51):
that
is, /' is the distance traveled by the impulse during time t'.
Generated on 2012-03-09 21:31 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] STEINMETZ: IMPULSE CURRENTS 17 In hyperbolic functions, equations (60) may be written: i =
1916] STEINMETZ: IMPULSE CURRENTS 17
In
hyperbolic functions, equations (60) may be written:
i
= B e-« |e-'T('-:'«> cosh hi ± e+s «-'•> cosh h (I - /')(
e
= B V% e-"< {€-*" sinh h I ± e+« sinh A (J - /'))
( (62)
C. PERIODIC IMPULSE CURRENTS
Substituting (37) and (38) in (4) and (6), separating the
imaginary exponentials from the real ones, substituting the
trigonometric expressions for the former, and rearranging, gives
i
= (-" {e-sl [Di cos kl - DI sin kl\ + f " [D, cos
*/-
Z>4sin*/]l
~r
(-"' \c e+s' [Dt cos>/ + D^ sin */] + - «~s/
o
c
[Z>4 cos*/ + Z?3sin*/]( (63)
where
c
= \/J5L±_L (54)
m
— s
Substituting:
£>, = db B (-«' cos kli
ZJ2
= ± B e-3'' sin */,
Dt
= ± B e+"i cos jfc (/, - /0) (66)
Z?4
= ± B e+5'- sin fe (/, - /0)
c-
«+"• (56)
Thus,
log_c= 1 J«±£ (67)
S
25 WZ — 5
and
writing
/ for < - /i + to
I for / + /o
gives
i
= B e-« {«+»('-« cos W ± e-*»-« cos jfe (/ - /»)}
e
= 5 V - e— {e+" sin */ ± e-s' sin k (I - *,))
(58)
Exchanging sin and cos in (66), also exchanges sin and cos
in
(68).
Generated on 2012-03-09 21:31 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
IS STEINMETZ: IMPULSE CURRENTS [Jan. 14 Equations (68) of the periodic impulse have the same
IS
STEINMETZ: IMPULSE CURRENTS [Jan. 14
Equations (68) of the periodic impulse have the same form as
equations (52) of the non-periodic impulse, except that the trigo-
nometric functions of distance take in the periodic impulse the
same position as the hyperbolic function in the non-periodic
impulse.
Current and voltage are in quadrature with each other in their
distribution in space, in either of the two components of the
periodic impulse. That is, in each of the two components
maximum current coincides with zero voltage, and inversely.
The
two components of the periodic impulse differ in the phase
of
their space distribution by the distance /o, the second compo-
nent
lagging behind the first component by the distance lo-
in
each of the two components of the periodic impulse, the
current lags behind the voltage by the time /„.
Current and voltage thus are in quadrature with each other in
space, and displaced from each other in time, by the " time dis-
placement " 10.
/o
is positive, that is, the current lags behind the voltage by-
time
to, if m is positive, and to is negative, that is, the current
leads the voltage, if m is negative.
Since m = 1/2 (^- --£-) it follows:
f
I?
The
current lags behind the voltage, if —=— > -—• , that is, if
L*
C
the
resistance-inductance effect preponderates.
The
current leads the voltage if -£— >—-.—, that is, if theca-
C
' Li
pacity effect preponderates.
The
voltage equals the current times the surge impedance
,
but is in quadrature with it in space, and the cur-
-V-
C
rent
is lagging by 10 in time.
By
the conditions of existence of the periodic impulse, 5 must
numerically be smaller than m.
5 = 0 gives
by
(57): st0 = 0.
by
(37): k = m \/L C
Generated on 2012-03-09 21:31 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] STE1NMETZ: IMPULSE CURRENTS 19 and by (68), i = B «-"< (cos k I'
1916] STE1NMETZ: IMPULSE CURRENTS 19
and
by (68),
i
= B «-"< (cos k I' =fc cos & (/' - /„)!
-g-
e-»'' {sink I' ± sin* (/'- /„)!
Hence, current and voltage are in phase in time, but in quadrature
in
space.
s
= m gives
k
= 0
hence, from (63),
e
=
hence, substituting for u and w,
- ±i --i
i = D! 6 ' + D3 € L
where
£>,' = Z>,
In
this impulse, the capacity terms and the inductance terms
are
separate, and current and voltage are uniform throughout the
entire circuit.
The
constants D or A or B are determined, as integration
constants, by the terminal conditions of the problem.
For
instance, if at the starting moment of the impulse, that is,
Generated on 2012-03-09 21:31 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
20 STEINMETZ: IMPULSE CURRENTS (Jan. 14 at time t = 0, the distribution of current
20 STEINMETZ: IMPULSE CURRENTS (Jan. 14
at
time t = 0, the distribution of current and of voltage through-
out
the circuit are given, we have by (63), for t = 0.
i
= {£>i + Z?a) cos kl — (Dz + D^ sin kl}
e
= -~r
+
(c Dl + -^-^s
The
development of the given distribution of current and
voltage into a Fourier series thus gives in the coefficients of this
series the equations determining the constants D\, Dz, Dz, D±
The
expressions for i and e, given in equations (39), (44), (45),
(48), (50) and (52) for the non-periodic, and in equations (53)
and
(58) for the periodic impulse, obviously apply to a simple
impulse only, and a general impulse is represented by the sums
2"
i and 2 e of all the 'expressions i and e, whose integration con-
stants satisfy the terminal conditions of the problem.
Generated on 2012-03-09 21:31 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] DISCUSSION AT NEW YORK 681 DISCUSSION ON " OUTLINE OF THEORY OF IMPULSE CUR-
1916] DISCUSSION AT NEW YORK 681
DISCUSSION ON " OUTLINE OF THEORY OF IMPULSE CUR-
RENTS " (STEINMETZ), NEW YORK, JAN. 14, 1916. (SEE
PROCEEDINGS FOR JANUARY, 1916.)
(Subjicl to final rmsion for tht Transactions.)
Charles P. Steinmetz: " Outline of Theory of Impulse
Currents" is a continuation of the paper on the general equa-
tion
of the electric circuit, read eight years ago. In the previous
paper it was found that the general differential equation, which
applies to every electric circuit or section of circuit having
constant values of r, L, £, and C, can be integrated by an ex-
pression consisting of four terms, two main waves and their
two
return waves. One of the waves dies out at a greater,
the
other at a slower rate than corresponds to the energy dissi-
pation in the circuit, and therefore the former transfers energy to
the
latter, thus representing the energy transfer along the
circuit, as occurring in traveling waves, a-c. transmission etc.
These two waves coincide, and the energy transfer coefficient
becomes zero, in the stationary oscillation of a uniform circuit;
they
do not coincide, however, in the stationary oscillation of
a
compound circuit, but energy transfer occurs from sections
of
lower energy dissipation, to sections of higher energy dis-
sipation.
In
the first part of the present paper, a classification of the
different types of electric current is made from the general equa-
tion
of the electric circuit, and in the second part, various forms
of
the equations of the general impulse currents are derived.
Two
methods of studying engineering phenomena exist,
which may be denoted respectively as the synthetic method
and
the analytic method.
The
synthetic method starts with the study of special cases,
and
by correlation of the special cases, by generalization and
classification progresses from special to general, and -thus
finally to the complete structure of the engineering science.
The
analytic method starts from the general (differential)
equation of the science based on the fundamental underlying
laws,
and by substituting all the possible values of the con-
stants and the terminal conditions, thereby derives the differ-
ent
classes of the phenomena, thus progressing from the most
general to the special.
As
engineering is based on experience, and experiment neces-
sarily deals with special cases, the synthetic method is the first
in
the development of engineering, and the analytic method,
requiring the knowledge of the fundamental laws, can be at-
tempted only later.
However, the synthetic method can never give assurance
of
the completeness of our knowledge, for entire classes of
phenomena may be omitted, if it happens that they have
never been empirically observed or recognized. Inversely,
the
analytic method gives the complete structure, but only
so
far as it is based on the fundamental laws represented by
Generated on 2012-03-09 20:56 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
682 IMPULSE CURRENTS [Jan. 14 the general equation, and thus may not he comprehensive, since
682 IMPULSE CURRENTS [Jan. 14
the general equation, and thus may not he comprehensive, since
any phenomenon which does not obey these laws, would not
be included.
Thus the two methods are supplementary, the analytic
method checking the completeness of the synthetic, and the
synthetic checking the comprehensiveness of the analytic.
The theory of electrical engineering, relating to direct currents,
alternating currents, exponential discharges, oscillating currents
and other transients, was developed synthetically; and the
purpose of the first part of the present paper is, by the analytic
method, to derive all possible types of currents by substituting
all possible values of constants and of terminal conditions,
and thereby check, whether any class of current of indus-
trial importance has escaped recognition. This appears prob-
able, as the observed transmission line phenomena do not
entirely agree with the characteristics of the currents by which
they are usually explained.
During the development of high-voltage long-distance trans-
mission, phenomena were observed during disturbances in the
transmission lines or underground cables, which could not be
explained by the normal voltage supplied by the generating
machinery. The first attempt at explanation was made in
the study of the "natural period" of the circuit, the abnormal
voltages resulting from the free oscillation of the line as a quarter
wave (or half wave or full wave), and its higher harmonics.
In a few instances this agreed fairly well with the facts, thus
pointing to quarter-wave oscillations as possible line disturbances.
Usually, howevei, it did not agree for a quarter wave oscillation
would be felt over the entire circuit, whereas, experience showed
most line disturbances were more local in chaiacter, that is, very
severe at some places, but rapidly decreasing with the distance. It
further showed, as characteristic, the piling up of voltage locally,
especially at inductive parts of the circuit, such as transformer
end turns, inductances, current transformers etc. This led to the
explanation of the disturbance as due to high frequency. High-
frequency travelling waves would give local abnormal voltage,
and rapid attenuation with the distance from the origin, and
•"herefore would .satisfactorily explain the most frequent line dis-
turbances, except in one feature, namely, that such high-frequency
oscillating currents should give pronounced resonance effects.
Such resonance effects, leading to the formation of stationary os-
cillations of destructive value, have been observed and experi-
mentally reproduced in recent years, in the high-potential wind-
ings of high-voltage power transformers, usually of frequencies
between 10,000 and 100,000 cycles, and their existence has there-
fore been proved. However, in most cases of transmission line
disturbances, resonance phenomena are remarkably weak or
entirely absent, and it therefore appears that many trans-
mission line disturbances are impulsive rather than oscillatory,
which has led to the question of the existence, the characteristics
and the equations of impulse currents.
Generated on 2012-03-09 20:57 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] DISCUSSION AT NEW YORK 683 In a circuit with localized capacity, inductance and resistance,
1916] DISCUSSION AT NEW YORK 683
In
a circuit with localized capacity, inductance and resistance,
the
current is either oscillatory or impulsive, depending on the
circuit constants, and more particularly the relation of the
resistance to the inductance and capacity. In a circuit with
distributed constants, however, there is no critical value of
constants, that divides the oscillatory and the impulsive
phenomena, but, as was shown in the paper of 1908, with the
same circuit constants, the phenomena may be oscillatory or they
may
be impulsive, depending on the terminal conditions, that
is,
on the cause of the phenomena. Thus oscillating currents
as
well as impulse currents may exist in the same circuit, al-
though experience seems to show, that at least in long distance
transmission lines the latter are rather the more frequent.
To
get their relation to the other and better known classes
of
current, was the purpose of the first part of the present paper.
1.
Terminal conditions. The foremost terminal condition is
the
length / of the circuit. This may be either zero, or finite,
or
infinite. Substituting / = 0 gives the equation of cir-
cuits
with massed constants: the usual a-c. or d-c. circuits of
our
systems and apparatus, etc. / = finite gives the equation
of
circuits with distributed constants and / = oo gives the
case
where the circuit is so long, that the disturbance has de-
creased to a negligible value before reaching the end of the
circuit, and thus the reflected disturbance is inappreciable.
2.
Constants. The general integral equation of the electric
circuit appears as an exponential function, with the time and
the
distance in the exponent. It thus has a coefficient of the
time
exponent, b, and a coefficient of the distance exponent, a.
a
and b are related by a quadratic equation, thus only one is
independent, b has been chosen as the independent co-
efficient, b is a general (or complex imaginary) number, and the
two
main special cases thus are, (a) where the real term of b
is
zero, (b) where the imaginary term is zero, (a) gives the al-
ternating currents, (b) gives a non-periodic class of transient
currents, which may be called the impulse currents. The im-
pulse currents thus appear as a class of currents, as general as,
and
coordinate with the alternating currents, the latter rep-
resenting the useful currents of our transmission systems, the
impulse currents the foremost type of harmful currents.
The
second part of the present paper contains a further
classification of the impulse currents, by their distribution
along the circuit, as non-periodic and periodic in space, and a
derivation of various forms of the equations of the two classes
of
impulse currents.
Physically, impulse currents, by the steepness of their wave
front, give the local piling up of voltage, characteristic of most
line
disturbances, but as non-periodic currents, they could
give
resonance phenomena only by multiple reflection, and thus
resonance phenomena with impulse currents would be little
pronounced or absent.
Generated on 2012-03-09 20:57 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
684 IMPULSE CURRENTS [Jan. 14 It appears, therefore, that the most frequent disturbances of our
684 IMPULSE CURRENTS [Jan. 14
It appears, therefore, that the most frequent disturbances
of our transmission systems show the characteristics of the
impulse currents rather than those of oscillating currents, and
the study of impulse currents becomes of far greater importance
than heretofore assumed.
Thus, the analytic study led to the recognition of the impulse
currents as a class of currents, which, while not unknown be-
fore, but repeatedly mentioned and discussed, apparently has
not sufficiently been realized in respect to its industrial im-
portance.
The reverse operation would now be of interest: to check
by the synthetic method, the completeness of the analytic
structure, that is, to see to what extent existing or at least
industrially important classes of currents are not contained within
the scope of the general equation based on constancy of r,
L, g and C.
Phenomena are known, which are outside of this equation.
Such phenomena are the cumulative oscillations, such as are
produced under certain conditions by an arc (not the phe-
nomena of the so-called 'arcing ground;' these are recurrent
oscillating discharges), the surging of synchronous machines,
the phenomena in circuits operating above corona voltage, etc.
Thus the general equation of the electric circuit, which is
the starting point of the present paper and the previous paper,
is not all comprehensive, but a still more general analytical
investigation is desirable, in which r, L, g and C are not con-
stant, but depend on — and -7— or an integrated value there-
Qfl Qflr
of, as the frequency, etc. Our knowledge of these phenomena
is not yet sufficient to attempt an analytic treatment, but
more knowledge will have to be acquired by the synthetic
method of investigating special classes of phenomena, in away
similar to that attempted with the surging of synchronous
machines in the paper on "Instability of Electric Circuits"
read before the Chicago Section in 1912.
M. I. Pupin: If I understand Dr. Steinmetz, the object
of his paper is to call your attention to a distinct class of elec-
tric currents, a class of electric currents which he calls impulse
currents and which, he says, has not received as much attention
as the direct current and the alternating current. When I saw
the notice of the paper and observed the title "Outline of Theory
of Impulse Currents," it attracted me very much, because I
have always been interested in impulse currents. To me the
direct current and the alternating current were simply cases
of the more general impulse current.
When Dr. Steinmetz says that equations (4) and (6) "must
represent every existing electric circuit, and every circuit
which can be imagined, from the lightning discharge to the
house bell, and from the a-c. transmission line to the telephone
Generated on 2012-03-09 20:57 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] DISCUSSION AT NEW YORK 685 circuit, with the only limitation, that r, g, L,
1916] DISCUSSION AT NEW YORK 685
circuit, with the only limitation, that r, g, L, C are constant
within the range of the currents and voltage considered," he
should not be misunderstood in his statement, and I believe
that he might be misunderstood. I wish to warn you not to
misunderstand that statement,"because if you do, you might
think that after you read this article you need not read any-
thing else on this subject, so, therefore, I want to warn ybu
against that.
Here Dr. Steinmetz considers the general problem on a
long line having distributed inductance, resistance and capacity,
and he gives you that, starting from a differential equation.
What do we mean by differential equations in electricity?
We mean simply this—an equation expressing the various
relations between the reactions in a conductor. For instance,
take the first equation Dr. Steinmetz gives, which looks so
mathematical—as a matter of fact it is nothing more nor less
than the expression that in any element of conductor the sum
of the electrical actions is equal to the sum of the electrical
reactions. That is what he says, and that is Newton's third
law of motion, that the sum of action is equal to the sum of
reactions in every system of bodies.
Dr. Steinmetz says—Suppose there is that relation between
these various electrical reactions and various electrical actions,
then the following must be the relation between the current and
e.m.f. in any part of the circuit. That is what is called the in-
tegral of it. From the equation of reactions you get an ex-
pression of the current and the e.m.f., and that is called the
integral. That is true for any point of the wire which has
certain constants.
Now, when you come to another point of the wire where
other constants are, then you have to get another expression
for the currents, and since you have an infinite number of
points, you may have an infinite number of different expressions
for the current, and it is necessary to add these different currents
and make them conform to the so-called boundary conditions.
As soon as you pass from one element of the wire with certain
constants, to another element of the wire with another set of
constants, you have to pass through that boundary.
The most difficult thing in mathematical analyses of elec-
trical phenomena is that question of the boundary conditions.
So that when Dr. Steinmetz says "These equations must rep-
resent every existing electric circuit," he does not mean to
say that he has given you a complete solution—-he means to
say that he has given you a complete solution for any part
of the electric circuit, but if you want to have a complete solu-
tion, good for any part and for the complete circuit, you have
to take this part, and this part, and this part and build it up.
That explains the point I had in mind.
It is true that there are problems in electrical engineering
which have not been discussed at all, and Dr. Steinmetz has
Generated on 2012-03-09 20:57 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
680 IMPULSE CURRENTS [Jan. 14 referred to one of those problems which has interested me
680 IMPULSE CURRENTS [Jan. 14
referred to one of those problems which has interested me for
years, the problem of the oscillating arc. It may be that in
some oscillating arcs the oscillation is due to the variable
resistance. It may be. But the electrical oscillation that
takes place in a vacuum tube, such as the pliotron tube, is not
due to variable resistance, but is due to something else. That
is to say, the oscillating arc acts in a very similar manner to
the oscillations produced by an induction motor when you
drive it beyond synchronism. Take a single-phase induction
motor, drive it beyond synchronism, and provide it with a suit-
able capacity, and you can get oscillations exactly the same
as you do in an oscillating wire in the pliotron tube—there is
no variable resistance, there is no variable capacity, there
is no variable inductance, the only thing that is variable
is the mutual inductance between the primary and secondary
circuit.
Dr. Steinmetz to my knowledge has not attempted yet to
proceed analytically and pursue this elusive induction motor
to see what it will do under certain conditions, but un-
doubtedly he will, and when he does he will find that an
induction motor, whether it is single-phase or polyphase, when
supplied with proper capacity and constructed suitably can
generate these oscillations in just the same way as the pliotron
tube or the oscillating arc. Moreover, if he does not take the
proper precaution, the oscillations stored in an induction motor
of that kind will be oscillations, not with a negative exponent,
but with a positive exponent; that is to say, the oscillations will
go on increasing indefinitely until his machine is smashed.
The machine has to obey the integral of that differential equation,
that is to say, the machine has to obey the law of the electrical
reactions. The mechanical power that drives the motor acts,
the motor reacts, and the result of that action and reaction is
continually piling up energy which appears as magnetic energy,
and when the current is big enough, of course, your machine
will be either smashed, or the pole pieces will be crushed on to
the armature and the machine will come to a standstill.
Harold Fender: Dr. Steinmetz has given us an interesting
mathematical discussion of an important class of electric pheno-
mena. It is not difficult to see the physical meaning of the
mathematical symbols used in the differential equations given
in the paper, but this is not true of many of the symbols appear-
ing in the integral equations. For example, on the last page of
the paper there are certain constants enumerated, namely,
DI, Z)j, Dz, D4. What are they? In mathematical language
they are called integration constants, and physically they
have a definite relation to the physical conditions intially
imposed on the circuit, i.e., they depend upon the voltage and
current initially established at each point of the circuit. But
how may these constants be evaluated in terms of these initial
conditions? I hope that Dr. Steinmetz will give us at some future
Generated on 2012-03-09 20:58 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] 687 DISCUSSION AT NEW YORK time a discussion not only of the qualitative meaning
1916]
687
DISCUSSION AT NEW YORK
time a discussion not only of the qualitative meaning of these
constants, which is not difficult to see, but also their quantitative
values for various initial conditions which may occur in practise.
The differential equation (3) in this paper is one that contains,
as Dr. Steinmetz has said, a complete solution of single-circuit
lines. By a single-circuit line I mean a circuit which does not
contain mutual inductance. If there is mutual inductance a
second equation is necessary, and the evaluation of the exponents
in the various integral relations requires the solution of a cubic
equation instead of a quadratic. When there is mutual induc-
tance between the given circuit and more than one other circuit,
the evaluation of the exponents requires the solution of an equa-
tion of still higher degree. I give this merely to emphasize the
fact that the relations given in the paper apply only to a single
circuit which is not influenced by any neighboring circuit, a
condition which seldom obtains in any transmission system.
Hans Lippelt: The paper, after introducing equations (1)
and (2), puts forward the following statement:
" These equations must represent every existing electric
i
4
'Source of
Electricity
\ To Ground or Opposite Polarity
FIG. 1
circuit, and every circuit which can be imagined, from the
lightning discharge to the house bell, and from the a-c. trans-
mission line to the telephone circuit, with the only limitation,
that r, g, L, C are constant within the range of the currents and
voltages considered."
By a peculiar coincidence, it occurred to me that these equa-
tions might not include all cases which are situated between the
limits given in that statement. As a matter of fact, Dr. Stein-
metz in concluding the presentation of the paper tonight men-
tioned several cases which are not included in these equations.
He limited himself, however, again by saying that the cited cases
involve variations of the " constants."
To me it seems that there is still another possibility of an
electric circuit, which is not covered by equations (1) and (2).
The circuit I have in mind contains capacity in series connec-
tion.
Fig. 1 shows in a general way an elementary circuit to which
Dr. Steinmetz's equations (1) and (2) have reference. The
figure will be readily understood by observing the notations used
Generated on 2012-03-09 20:58 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
688 [Jan. 14 IMPULSE CURRENTS in the paper. The whole circuit is composed of a
688
[Jan. 14
IMPULSE CURRENTS
in
the paper. The whole circuit is composed of a series of such
elementary circuits.
If
we have a capacity C\ in series with the circuit, the latter
may
be represented as shown in Fig. 2. It is not feasible, how-
ever,
to assume capacity in each elementary circuit, because the
capacity of the total circuit would then be so small that no
current at all could flow. The new fundamental equation should,
therefore, not refer to the change of voltage per unit length of line
(~rr )• but to a voltage drop in a circuit with massed constants.
The
length of circuit as such does then not enter into the com-
putation.
If
e, the consumed voltage, and the other quantities r\, r2, LI,
Li,
Ci, Ci g, refer to the circuit as per Fig. 2, we have
e
=
dt,
(1)
Source of
Electricity
Leads assumed free
""from resistance
FIG. 2
Similarly, the current diverted from main circuit is, allowing
also
for self-induction L2 combined with resistance r2,
=
ge + C2 —f- + ik X t L' + -
at
wherein ik is defined by the following:
•r,' C +^«
—
edt+I"1
ft
I
(2)
**
e0
1 fe+r,'
—
to -) I c
r2
r, J
,
de
finally
i-o
const. -0
+
(3)
Generated on 2012-03-09 20:58 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] DISCUSSION AT NEW YORK 689 To solve equations (1) and (2) requires the knowledge
1916] DISCUSSION AT NEW YORK 689
To
solve equations (1) and (2) requires the knowledge of e
as
a function of the time, and in this respect these equations
differ
materially from equations (1) and (2) of the paper. The
latter refer to gradients of quantities [ —^- and —jr- \ , which are
independent of actual values of e and i. Equations (1) and (2)
refer
to the quantities e and i directly, of which e is equal to the
impressed voltage, which governs the whole process, viz., the
vector sum of all secondary e.m.fs. must equal e. In the case of
Fig. 1, forming only a small part of a large circuit, this partial
circuit may draw stored power from adjacent circuits, involving
an
adjustment of its terminal voltage, and therefore equations
(1) and (2) of the paper must leave this possibility open, which
they
do.
A
partial, or rather advanced solution of equation (1) is given
in
equation (4).
L.
with
x
,
= --£- +V- ri'
2L,
4Li2
--J3--A/:
2L,
4 L!* Ci Li
Kt
and K2 = integrating constants.
The
last two terms on the right-hand side of equation (2)
represent the current flowing through the branch loaded with
inductance L« and resistance r2, Fig. 2. These terms have been
found by treating this branch separately.
To
complete the solution of equation (2) requires only the
substitution of the supposedly known value of e (as a function
of
time) and carrying out a simple mathematical process.
An
application of a circuit with capacity is found in high-
voltage d-c. machines, having armature windings of the open
coil
type. Such machines work entirely with impulse currents
and
it appears that circuits as per Fig. 2, or similar, will meet the
requirements for sparkless commutation of current in the wind-
ings.
Generated on 2012-03-09 20:58 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
690 IMPULSE CURRENTS [Jan. 14 A. E. Kennelly (by letter): The great advantages of the
690 IMPULSE CURRENTS [Jan. 14
A. E. Kennelly (by letter): The great advantages of the
analytical method set forth in Dr. Steinmetz's valuable paper
are:
(1) The integral or primitive equations for all voltage and
current waves over conductors are reduced to their simplest
fundamental elements.
(2) These equations (4) and (6) of the paper permit of a new
classification of all such voltage and current waves.
When b contains a real component, the paper shows that the
wave
to which it pertains must speedily disappear. Only when
b
has no real component can the wave to which it pertains belong
to
a permanent regime.
There are three cases involving real components of b; namely,
b
complex, representing oscillatory transients (i)
,
. j with a real, representing non-oscillatory transients (2)
r
j with a imaginary, representing non-oscillatory transients.
(3)
The
paper distinguishes the two last types by the terms " non-
periodic " and " periodic " impulses respectively. But these
terms seem to be unsuitable because they suggest recurrence in
time;
whereas the property in question is a recurrence in dis-
tance or space. Would it not therefore be more appropriate
to
coin the terms " non-spacic " and " spade " for this differen-
tiation?
In
any case, although the oscillatory and non-oscillatory tran-
sients should clearly be distinguished and placed in separate
categories, it seems doubtful whether there is enough distinction
between the two classes (2) and (3) of the non-oscillatory tran-
sients to make their separation important. The paper shows
that
the only essential difference between these two types is
that
where a circular distance-angle occurs with the one, a hyper-
bolic
distance-angle occurs correspondingly with the other.
Thus
both are included in the same generalized trigonometric
relations and it remains to be shown whether the differences
between them are otherwise great enough to call for separate
discussion.
It
is perhaps going too far to say that all impulses with real
b
exponential time-factors are harmful; although the generaliza-
tion
may at present be applied to light and power circuits. In
some signaling circuits, as in some submarine cable circuits,
such
impulses discharging back to ground at the sending end are
usefully employed in certain signaling systems.
Charles P. Steinmetz: In regard to the general equation,
which I gave in my paper eight years ago, naturally I did not
mean that this equation is a final solution of every phenomenon;
if
I did, I would not have had any reason to write this paper.
What
I mean to say is that from this equation can be derived
the
equations of any circuit which fulfills the condition that every
one
of these four constants r, %, L and C is constant. Where one
is
variable, whether resistance, or inductance, or the capacity
Generated on 2012-03-09 20:59 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google
1916] DISCUSSION AT NEW YORK 691 or conductance, then this general equation does not apply.
1916] DISCUSSION AT NEW YORK 691
or
conductance, then this general equation does not apply.
I
mentioned, as an illustration, a couple of circuits where the
constants are variable. The transmission line, when operated
above corona voltage, also gives a periodically variable con-
ductance.
I
referred to the oscillating arc. There I referred to the variable
resistance. In the arc it is really a variable effective resistance,
as
I may call it, defining, as I have done here, resistance as the
coefficient of energy dissipation proportionate to the current.
The
induction machine driven above synchronism also produces
oscillations. These oscillations produced by the induction
machine when driven above synchronism were discussed in my
paper on "Induction Motors" 19 years ago, and their characteris-
curves calculated. In the chapter on " Induction Generator "
in
" Alternating Current Phenomena," a full discussion is
found on the conditions under which an induction machine
above synchronism excites, of the maximum voltage and cur-
rent
values to which it may build up, and its dependency on
the
constants of the external circuit. It may be interesting to
note
that two such induction, generators of 10,000 kw". each,
have been in operation for years in the Interborough Rapid
Transit station in New York City, as generators producing
power.
These impulse currents are a special class of transients.
Inasmuch as they are a sub-class of the general transient, they
are
included in the general equation of my previous paper, but
they
were not specifically treated.
They apply to a circuit, or a section of the circuit, of uniform
constants, but the case which Prof. Pupin discussed, where the
circuit constants change, is treated in general in my previous
paper under the term " Complex Circuits." It is more particu-
larly
treated in that section in my book on " Transient Phe-
nomena," which discusses the transition points between different
circuit sections of different constants, as lines, transformers, etc.
It
is very interesting to note the effect on such a circuit of the
transients existing; there is an energy exchange between the differ-
ent
circuit sections, the energy being dissipated in some sections
at
a rate higher than the average, and in other sections at a rate
lower than the average,—there is energy transferred, taken
from one section and delivered at another section.
With
regard to mutally inductive circuits, even these may be
considered under the genera! equation, by suitable terminal
conditions, and effective values of r, g, L, C.
I
may say that the circuit as described by Mr. Lippelt is
in
industrial existence. It is the circuit of the multi-gap light-
ning
arrester.
Generated on 2012-03-09 20:59 GMT / http://hdl.handle.net/2027/mdp.39015075024300
Public Domain, Google-digitized / http://www.hathitrust.org/access_use#pd-google